27 Mar
2009
27 Mar
'09
4:09 p.m.
On Fri, 27 Mar 2009 10:21:28 -0700
Fernando Lopez-Lezcano <nando(a)ccrma.Stanford.EDU> wrote:
On Thu, 2009-03-26 at 22:20 +0100, Robin Gareus
wrote:
Yes, no hardware problems AFAICT.
-- Fernando
After today I'd say I do have a lot more lockups than I
had with my distro's default preempt but non-rt 2.6.28 kernel.
Also have yet to try vanilla 2.6.29.
I'd also like to have a look at your config if possible, I hear good
things about your kernels.
Best regards,
Philipp
Fernando Lopez-Lezcano wrote:
I have not tested with vanilla 2.6.29 but the problem happens in
several very different machines. I have sent feedback to
lkml/Thomas/Ingo.
On Thu, 2009-03-26 at 19:52 +0100, Robin Gareus
wrote:
We've got a winner! 2.6.29-rt1
This is related to your BIOS or motherboard. Do you have the same
problem with a vanilla 2.6.29? Hmmmm, getting close but not yet a winner :-)
> I've been tracking the 2.6.29 rcx series and I'm still having
> some problems. 2.6.29-rt1 still does not shutdown correctly with
> my kernel configuration options (you need to specify
> "noreplace-smp" in the kernel startup line for shutdown to
> happen - otherwise you get a problem at the very end of the halt
> and the machine does not power down). Since I only did reboots, I just performed a
shutdown to check and
it powers off just fine. I don't use any boot options (only
"root=xx quiet ro") on this dual-core intel.
Why shut-down, anyway? s2ram and s2disk work fine with 2.6.29-rt1.
I'm very very happy (-:
Ok, I'll never turn off the machine and thus conclude there is no
bug! :-)
> And I
have seen it hung (hard) the whole machine where a normal
> kernel would just work. I'm still trying to find a way to debug
> that.
I did not have those for a long time (~2years), previously I
experienced hard lockups when using wifi (iwl3945) and rtirq.
I dare say there's no way to debug that - but you should ask that on
rt-linux list. There's ways to tweak things
in /proc/sys/kernel/sched_* if this lock-up is caused by a runaway
process.
It could also be CPU overheating, but then the problem should also
occur with non realtime kernels as well.